Cavitation pulse extraction and centrifugal pump analysis

This study extracted cavitation pulses from hydrophone signals sampled in a centrifugal pump and analyzed their characteristics. The modified and simplified Empirical mode decomposition (EMD) algorithm was proposed for extracting cavitation pulses from strong background noise. Experimental results showed that EMD can effectively suppress noise and obtain clear cavitation pulses, facilitating the identification of the number of pulses associated with the degree of cavitation. The cavitation characteristics were modeled to predict the value of incipient cavitation. Then, we proposed a method for detecting the wear of the impeller surface. That is, the information on the impeller surface of the centrifugal pump, including the roughness of the impeller surface and its wear trends, were quantified based on the net positive suction head available of incipient cavitation. The findings indicate that the proposed technique is suitable for condition monitoring of the pump.